The purpose of this project is to conduct statistical research and provide consultation to the Division for clinical trials, laboratory experiments, and epidemiological studies relevant to cancer prevention and control. Research problems under investigation include statistical methods for validating intermediate endpoints for cancer research, where the statistical criteria for validation are being elucidated and their application to biomarker research are being described; design of nutritional cohort studies including two-stage designs that allow construction of a followup cohort with a greater variation of nutrient intake; design of studies to validate dietary assessment instruments, allowing repeated assessments of two different instruments, one of which is unbiased; use of Bayesian methods for monitoring clinical trials; studies on a large observational database of HIV-infected subjects with evaluation of natural history of disease; the use of caloric adjustment models in the analysis of nutritional epidemiology studies, with emphasis firstly on the interpretation of three commonly used models when nutrient intakes are expressed as continuous variables and secondly on the interpretation of the same models when nutrient intakes are categorized as belonging to a certain quantile of the population distribution. In the latter case the results show that categorization leads to surprising differences with higher relative risks obtained from the "standard" model than from the "Willett" model. Statistical consultation is provided to numerous studies including the NIH Women's Health Initiative Clinical Trial, the Polyp Prevention Trial, the AARP Nutritional Cohort Study, and several projects of the DCPC Chemoprevention Program. The consultation has involved extensive contributions to study design, reviewing proposals for interpreting the studies and, in the case of the Polyp Prevention Trial, continual advice on the day-to-day operations and on data monitoring.